How private ADHD clinics say they diagnose

Chapter one

How fast is the private ADHD market growing, and how much of it closes?

146 of the 272 companies behind listed private ADHD clinics were incorporated in 2020 or later, and 52 of them in 2024 or later. On the CQC register, 37 locations belonging to these clinics registered during 2025 and 10 closed; in 2026 to 18 September 2026, 20 registered and 14 closed.

The oldest company in this set was incorporated in 1979. The median one was incorporated in 2020. That is the shape of a sector that assembled itself in about five years, and it explains a good deal of what follows: a young company has not had an inspection cycle, has no long record to publish, and is often one clinician with a website.

Closures are the part of a young market that is easy to read wrongly. A location leaving the CQC register is not necessarily a clinic failing: 7 clinics both closed one location and registered another during 2026, which is a relocation. What the register does show is a short average tenure. The median currently registered location has been registered for 2.4 years; the median closed one lasted 2.1 years, and 26 of 59 closed locations lasted under two.

How to check this
Incorporation dates are the date_of_creation column of companies_house.csv, each row carrying the Companies House URL it was read from. Registrations and closures are registration_date and deregistration_date in cqc_register.csv, each with a location_url on cqc.org.uk.

Chapter two

Who is behind the clinics?

265 of 342 listed clinics have a UK company behind them; 77 have none on record, which means a sole trader, a partnership, an NHS body or an overseas entity. Of the 272 companies, 40 list a corporate entity as a person with significant control, and 2 of those parents sit above more than one clinic in this directory.

This is a sector of small, separately incorporated businesses rather than a handful of chains. That has a direct consequence for a patient: there is no group-level complaints route, no shared quality framework and no common definition of what the assessment includes. It also means the register entry is often young — 54 of the companies have traded under a previous name.

How to check this
Company links are in companies_house.csv, with a link_basis column saying how each was established. Corporate parents are corporate_psc_names. No company, parent or director is named on this page; all of it is in the published file, which carries the Companies House URL for every row.

Chapter three

Who carries out a private ADHD assessment?

193 of 337 clinic websites (57%) name the professional role that carries out the assessment. On 84 sites a psychiatrist is named; on 48 a psychologist is named and no doctor is; on 5 a paediatrician and no psychiatrist; on 15 a nurse or pharmacist and neither a doctor nor a psychologist; and on 41 the only description is an unspecified “clinician” or “specialist”. The remaining 144 name no role at all.

NICE is deliberately broad here. Recommendation 1.3.1 of NG87 says ADHD should only be diagnosed by “a specialist psychiatrist, paediatrician or other appropriately qualified healthcare professional with training and expertise in diagnosing ADHD”. A specialist nurse with the right training and expertise falls inside that; so does a clinical psychologist. None of the groups below is non-compliant by virtue of its job title, and this chart is not a quality ranking. What it shows is which sites let a patient know in advance, and 144 of them do not.

Counting every role a site mentions rather than one per site gives the vocabulary of the sector. The single commonest term is not a profession at all.

A related question is who takes responsibility for the diagnosis once it is made. 47 of 337 sites say who signs it off — the least-published fact of the 20 we looked for.

How to check this
Column assessing_clinician_types in assessment_procedures.csv, with assessing_clinician_types_quote holding the sentence it came from and assessing_clinician_types_source_url the page. Sign-off is diagnosis_signed_off_by, stored as the site’s own words.

Chapter four

What does an assessment contain, measured against NICE NG87?

NICE recommendation 1.3.1 describes a diagnosis made on a full clinical and psychosocial assessment, a full developmental and psychiatric history, and observer reports. 74 of 337 sites publish all three of those elements — naming who assesses, describing a developmental history, and asking for an informant account. On the adult pathway alone it is 63 of 297.

Taken one at a time, the elements are published at very different rates. 155 sites (46%) describe taking a developmental history. 142 (42%) ask for an account from someone who knows the person — a partner, parent or old school report — and 11 of those say it is required rather than helpful. On the adult pathway, 7 sites require one. NG87 lists observer reports as part of the assessment, and for an adult recalling childhood symptoms it is often the only corroboration available.

Stacking the three core elements gives the distribution below. 84 sites publish none of them and 74 publish all three.

Claiming the guideline and publishing it

161 of 337 sites state that they follow NICE guidance. That claim does correlate with publishing more: 96 of 161 claimers describe a developmental history, against 59 of 176 of the rest. But the claim is not a substitute for the detail: 46 of the 161 sites invoking NICE publish all three core elements. A reader who treats “NICE-compliant” as a summary of the process is taking the claim on trust.

Instruments, questionnaires and computerised tests

88 sites name a structured interview or rating scale, most often a version of the DIVA diagnostic interview (65 sites) or the Conners scales (21). NG87 recommendation 1.3.2 is explicit that a diagnosis should not be made solely on rating scale or observational data, and describes scales as helpful added tools. Naming one is a sign of a structured process; it is not a process on its own.

75 sites mention a computerised attention test, 68 of them a QbTest or QbCheck (40 as a routine part of the assessment, 28 as an option). This is worth reading carefully. NICE HealthTech guidance HTG729 recommends QbTest as an option to help diagnose ADHD in people aged 6 to 17, alongside a standard clinical assessment. For adults, the committee was unable to make a recommendation, citing limited evidence, and said the evidence from under-18s does not generalise. 52 of the sites mentioning one do so on a pathway that covers adults. That is not a finding that they are wrong to use it — a test outside a NICE recommendation is not a prohibited test — but a patient told an assessment includes objective testing may reasonably assume a settled evidence base behind it.

How long it takes

112 of 337 sites state how long the assessment appointment lasts. Among the 106 stating a comparable figure, the median is 120 minutes and the range runs from 50 to 210. 4 state an hour or less; 9 state three hours or more. Appointment length is a description of the session, not of the whole pathway: a short appointment may follow a long questionnaire, and a long one may be the entire process.

35 sites say how many appointments the assessment takes. Where they do, one and two appointments are close to equally common.

How to check this
Each element is one column of assessment_procedures.csv, paired with a _quote and a _source_url. The child-pathway exclusion is the fields_stated_for_child_pathway_only column. NICE NG87 and HTG729 are linked in full under Sources.

Chapter five

Do clinics say the outcome comes on the day, and will a GP take over?

28 of 337 sites say the diagnostic outcome is given at the end of the appointment. 21 of those qualify it — usually “in most cases” or “where possible” — and 7 state it without a qualifier. 5 say the outcome comes after the appointment, and 304 do not address it.

Shared care is the arrangement under which an NHS GP takes over prescribing after a private diagnosis, at NHS cost to the patient. It is voluntary for the GP and many Integrated Care Boards restrict it, so whether a clinic can deliver it is not wholly in the clinic’s hands. The wording reflects that: of the 109 sites that mention shared care at all, 101 make it conditional on the GP agreeing or describe it as something they aim for, 6 word it as an offer, and 2 say it is not available. The remaining 228 sites do not mention it — a silence that bears directly on whether a patient’s costs end with the diagnosis.

On prescribing itself, 145 sites say they prescribe after diagnosis and 42 say they do not; 113 describe a titration or dose-adjustment service. 33 sites say they accept NHS Right to Choose referrals. Right to Choose is an England-only pathway, it needs a GP referral, and an Integrated Care Board may restrict or pause it.

How to check this
Columns same_day_diagnosis_claim, shared_care_stated, post_diagnosis_prescribing_stated, titration_stated and right_to_choose_stated, each with its quote. The corrections file records every shared-care value that was downgraded from an offer to a conditional reading during review.

Chapter six

Who checks private ADHD clinics?

Of 304 clinics within CQC’s remit, 132 match a CQC record and 128 have a currently registered location. Of the 142 currently registered locations, 40 hold one of CQC’s four quality judgements and 102 hold none.

Among registered locations, the striking figure is how few have been rated. Registration means a provider has satisfied CQC that it meets the requirements to operate; a rating follows an inspection, and inspections are prioritised by risk and capacity. The unrated locations are mostly recent arrivals: 69 of the 102 registered in 2024 or later. Of the 40 judgements that do exist, 34 are Good, 2 Outstanding and 4 Requires improvement. 15 of the ratings in the file were published under CQC’s single assessment framework; the most recent inspection under the previous framework anywhere in this set was 18 October 2023.

54 sites state on their own pages that they are CQC registered. Our matching corroborates 46 of those against a currently registered location and does not corroborate 7. For the reasons above, a non-corroboration is a statement about our matching rather than a finding that the claim is untrue.

Outside England, 38 listed clinics have all their locations in Scotland, Wales or Northern Ireland, where CQC has no remit. 12 appear on the Healthcare Improvement Scotland register of independent healthcare services, 10 of them registered and 2 shown as in progress. Healthcare Inspectorate Wales and the RQIA registers are not machine readable and were not pulled.

How to check this
cqc_register.csv carries one row per matched location with match_method, match_confidence, the rating, the inspection date and a location_url that opens the CQC record. cqc_match_status in clinics.csv says why any clinic is unmatched. Scotland is scotland_his_register.csv.

Chapter seven

What does the price buy?

Across 162 clinics publishing an adult assessment price, the median was £995, from £275 to £2,095. What the price tracks is not the amount of published process. It tracks who the clinic says will see you.

Delivery format moves the price in the direction you would expect, and by a wide margin. Clinics describing an online-only assessment published a median of £750; those describing an in-person-only assessment, £1,445. Company age does not move it: clinics whose company was incorporated in 2023 or later published a median of £999, against £962 for those incorporated before 2018. The group with no company on record — sole traders and partnerships — published the highest median of the four, at £1,238.

Clinics matched to a CQC record publish lower prices than unmatched ones, and the gap survives splitting the sample by whether the clinic says it prescribes. That rules out the simplest explanation — that unmatched clinics are assessment-only services charging for a one-off report — without establishing what does explain it. Both cells below are small and both overlap, so read this as a pattern worth checking rather than a result.

What almost does not move the price is the thing a patient might assume sets it: how long the appointment is. 61 clinics publish both a comparable length and a price. Those stating under two hours published a median of £772 (32 clinics); those stating two hours or more, £1,000 (29). At every length, the spread between clinics is far wider than the difference between lengths.

How to check this
Prices are the adult_assessment_price_gbp column of clinics.csv, each with the page it was read from and the date it was verified. They were checked in a dedicated sweep in July 2026 and cover one route per clinic — a complete adult diagnostic pathway, not an initial or screening appointment.

Chapter eight

How much can a patient learn before paying?

Counting how many of the 20 facts in this study each site publishes, the median clinic publishes 7. 34 of 337 sites publish none of them on the pages we read. 87 publish ten or more, and the highest published 18.

This is the summary measure the rest of the report builds to. It is a count of published facts, not a quality score: a clinic can publish nothing and assess impeccably. But a patient choosing between two clinics can only choose on what is published, and on a median site there are 13 things they cannot find out without asking.

Two groups publish consistently more. Clinics matched to a CQC record publish a median of 9 facts against 6 for unmatched ones, and clinics that publish a price publish 8 against 6. Neither is causal. The plausible common cause is simply the age and size of the organisation: an established provider has more to say and more staff to say it.

How to check this
The score is simply the count of non-empty, non-“not stated” values across the 20 fields of assessment_procedures.csv for that clinic. You can recompute it from the published file in a spreadsheet.

How this was done

What was read

Up to six pages per clinic website — the assessment, pricing, process, team, FAQ and about pages where they exist — 1,759 pages in all, collected in the week to 18 September 2026. 5 of 342 sites could not be read at all and are excluded from every denominator, which is why the base is 337.

How the values were extracted

A language model read the stored page text and proposed a value for each of the 20 fields, with a verbatim quote. A value without a quote that matched the stored page was discarded, not kept. The extraction proposed 2,561 values; 2,259 survived review and are published, every one with its quote and source URL.

How it was reviewed

A second pass re-read the quote alone and could only weaken a value, never strengthen it: an offer could become a conditional, a “yes” could become “not stated”, and the reverse was not permitted. A hand pass then worked through the cases the second pass flagged. Every change is in corrections.csv with its before, after, reason and which pass made it — 464 corrections in total. Publishing the corrections rather than only the corrected file is the point: it is how a reader can audit the direction of our editing.

Registers

Companies House and the CQC syndication API, both pulled on 18 September 2026; Healthcare Improvement Scotland from its published register. Clinics were matched to CQC by location identifier, or by postcode plus a distinctive shared name token. Weak matches were rejected rather than guessed, which makes the matched count a floor.

Links

Every cited URL in the published files was opened again after the files were final, and the result recorded. A handful of clinic sites refuse automated requests and were checked by eye in a browser instead; the data dictionary names them.

Reproducibility

Every number on this page is generated from the published files by a script in our repository, not typed by hand, and cross-checked against a second independent implementation of the same counts before the page builds. If you recompute a figure from the files below and get a different answer, we would like to know.

What this study cannot tell you

• It measures publication, not practice. Every figure here is about what a clinic says on its website. None of it is evidence about the quality of an assessment, and none of it should be read as a clinical judgement. We list clinics; we do not rate them.
• Every count is a floor. We read up to six pages per site. A fact stated in a downloadable brochure, a video or a page we did not reach is recorded as not stated.
• Websites change. This is a snapshot taken in the week to 18 September 2026. Some of these pages will already have been rewritten.
• Matching is conservative. Unmatched to CQC means we could not corroborate a record, not that none exists. The register itself lists what exists now, so closures well in the past are undercounted.
• It is a census of a directory, not of the market. These are the 342 clinics listed on this site. There is no public register of private ADHD providers in the UK, so no one can enumerate the whole market, and we make no claim to have done so.
• One price route per clinic. Prices are a complete adult diagnostic pathway where one is published. Clinics with tiered or bundled pricing are represented by a single figure.

Work this builds on

This is not the first examination of the private ADHD sector, and it deliberately does not repeat the parts others have done better. The Centre for Health and the Public Interest reported in January 2026 on the NHS-funded end of the market, using freedom of information requests to Integrated Care Boards — a route into spending and contracting that no website study can reach. NHS England’s ADHD taskforce reported in November 2025 on demand and capacity across the system. A 2024 survey by a health insurance comparison site collected quoted prices by asking clinics directly, where this study records only what a clinic publishes itself; quoted and published prices need not agree, and the two should not be read as a time series.

What is new here is the clinic-level detail on published assessment procedure, tied row by row to the company and regulator records for the same clinic, with the quote and the URL attached to every value.

Our own earlier study of prices and provision across the UK private ADHD market covers what these assessments cost and how little of that is explained by geography.

Sources

• NICE guideline NG87, Attention deficit hyperactivity disorder: diagnosis and management (recommendations 1.3.1 and 1.3.2)
• NICE HealthTech guidance HTG729, Digital technologies for assessing ADHD
• Care Quality Commission, public register of providers and locations
• Companies House public data
• Healthcare Improvement Scotland, register of independent healthcare services
• Centre for Health and the Public Interest, Market failure: how the under-regulated market in NHS funded ADHD services impacts patients and the finances of the NHS (January 2026)
• NHS England, ADHD taskforce final report (November 2025)

Questions

Changes to this report

No corrections have been made since publication on 19 September 2026. Any change to a figure will be listed here with its date and reason, and the published files will be regenerated at the same time.